Hydrographic changes of the Southern Ocean (southeast Indian Sector) Over the last 230 kyr

(1)

HAL Id: hal-02913419

https://hal.archives-ouvertes.fr/hal-02913419

Submitted on 6 Feb 2021

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Hydrographic changes of the Southern Ocean (southeast

Indian Sector) Over the last 230 kyr

Laurent Labeyrie, Monique Labracherie, Nabila Gorfti, Jean Jacques Pichon,

Maryline Vautravers, Maurice Arnold, Jean-Claude Duplessy, Martine

Paterne, Elizabeth Michel, Josette Duprat, et al.

To cite this version:

Laurent Labeyrie, Monique Labracherie, Nabila Gorfti, Jean Jacques Pichon, Maryline Vautravers,

et al.. Hydrographic changes of the Southern Ocean (southeast Indian Sector) Over the last 230

kyr. Paleoceanography, American Geophysical Union, 1996, 11 (1), pp.57-76. �10.1029/95PA02255�.

�hal-02913419�

(2)

PALEOCEANOGRAPHY, VOL. 11, NO. 1, PAGES 57-76, FEBRUARY 1996

Hydrographic changes of the Southern Ocean (southeast

Indian sector) over the last 230 kyr

Laurent

Labeyrie,

•'2

Monique

Labracherie,

3 Nabila

Gorfti,

• Jean

Jacques

Pichon,

3 Maryline

Vautravers,

3 Maurice

Arnold,

• Jean-Claude

Duplessy,

• Marfine

Paterne,

• Elizabeth

Michel,

• Josette

Duprat,

3 Michelle

Caralp,

3 and Jean-Louis

Turon

3

Abstract. Hydrographical changes of the southern Indian Ocean over the last 230 kyr, is reconstructed using a 17-m-long sediment core (MD 88 770; 46ø01'S 96ø28'E, 3290m). The oxygen and carbon isotopic composition of planktonic (N. pachyderma sinistra and G. bulloides) and benthic (Cibicidoides

wuellerstorfi, Epistominella exigua, and Melonis barleeanum) foraminifera have been analysed. Changes

in sea surface temperatures (SST) are calculated using diatom and foraminiferal transfer functions. A new core top calibration for the Southern Ocean allows an extension of the method developed in the North Atlantic to estimate paleosalinities (Duplessy et al., 1991). The age scale is built using accelerator mass

spectrometry

(AMS) 14C

dating

of N. pachyderma

s. for the last 35 kyr, and an astronomical

age scale

beyond. Changes in surface temperature and salinity clearly lead (by 3 to 7 kyr) deep water variations. Thus changes in deep water circulation are not the cause of the early response of the surface Southern Ocean to climatic changes. We suggest that the early warming and cooling of the Southern Ocean result from at least two processes acting in different orbital bands and latitudes: (1) seasonality modulated by obliquity affects the high-latitude ocean surface albedo (sea ice coverage) and heat transfer to and from the atmosphere; (2) low-latitude insolation modulated by precession influences directly the atmosphere dynamic and related precipitation/evaporation changes, which may significantly change heat transfer to the high southern latitudes, through their control on latitudinal distribution of the major frontal zones and on the conditions of intermediate and deep water formation.

1. Introduction

The Southern Ocean is a major component of the global climate system, in particular for its role in deep-water circula- tion and the oceanic carbon cycle [Toggweiler and Sarmiento, 1985]. Yet Southern Ocean hydrographic changes during glacial-interglacial cycles are still poorly known. It has been observed for some time that in Southern Ocean sediment cores, the changes in summer sea surface temperature (SSST) (recorded by the microfossils) lead the changes in northern hemisphere

ice sheets volume (recorded in the foraminiferal •80/160 ratio)

by as much as 5 kyr during the major glacial terminations [Hays et al., 1976; Climate: Long-Range Investigations, Mapping, and Prediction (CLIMAP), 1984; Howard and Prell, 1984, 1992; Labeyrie et al., 1986; Labracherie et al., 1989; Pichon et al.,

1992]. The mechanism linking this early response to the northern hemisphere climatic evolution is still not understood. It has been associated to the sensitivity of Southern Ocean surface water to changes in the input of North Atlantic Deep

•Centre des Faibles Radioactivit6s, Laboratoire mixte CNRS-

CEA, Gif/Yvette, France.

2Also

at D6partement

des Sciences

de la Terre, Universit6

Paris-sud Orsay, Orsay, France.

3D6partement

de G6ologie

et Oc6anographie,

URA 197 du

CNRS, Universit6 Bordeaux 1, Talence, France.

Paper number 95PA02255. 0883-8305/96/95PA-02255510.00

Water (NADW), itself modulated by Nordic Seas surface [Crowley, 1992; Imbrie et al., 1992]. Because changes in Southern Ocean surface hydrography apparently precede variations in all proxies which are known to influence NADW, Imbrie et al. [1992] proposed that some yet undiscovered process, occurring in the high northern latitudes, reacted faster

to northern insolation and influenced the southern waters.

Progress in this field is hampered by the lack of sediment cores representative of the Southern Ocean hydrography which can be analysed for both planktonic and benthic foraminifera. We present in this paper new data from core MD 88-770 (46ø01'S 96ø28'E, 3290 m). The core is located in the same general area as core RC 11-120 (43ø31'S 79ø52'E), a reference core for Southern Ocean paleoclimatology (Figure 1). The dis- cussion is mostly based upon the oxygen isotopic analysis of planktonic and benthic foraminifera and sea surface summer temperature (SSST)estimated using foraminiferal and diatom transfer functions. Sea surface salinity is derived from the

planktonic

foraminiferal

fi180

and SSST

following

Duplessy

et

al. [1991], with a calibration valid for the Southern Ocean. The

benthic foraminiferal

fi13C

further

constrains

the changes

in

deep water characteristics.

Two main results are derived:

1. Changes in SSST lead the response of Southern Ocean deep water, which indicates that changes in deep thermohaline circulation are not at the origin of the lead of the southern surface hydrography.

2. Sea surface salinities generally increased at the beginning of the surface water warming. This indicates a sudden invasion of subtropical waters from the north. Conversely, cooling

(3)

58 LABEYRIE ET AL.: SOUTHERN OCEAN PALEOHYDROGRAPHY

...

i

... -: ...

... ,

...

ß ':'!:-.'.'! ... ! ... ,: ... '•'4.-W-- 5 ... ! ... ' ... i ... '; ... ! ... i--.

...

....

'

'._

I}i• ;S. ubtro•cal Front i : •.: -,,,0•;. ... .

..•:.•

...

i•

...

i

...

i

...

t

...

i

...

• ...

i

.... !

...

i..MD88.-770..i...

•'•,•,,,o,,•

...

•o

o .•

...

i ...

• ...

i

...

,...

... i ... i ... •r ... 'r ... • ... i .... ,: ...

i i : : : ' '

70ø8

}

i

'

..i:i:i:i:i•'i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i:i

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

40 øW 0 40 ø 80 ø 1 20 øE

Figure 1. Distribution

of published

Southern

Ocean

cores

with benthic

8•80 records

covering

at least the

last 140 kyr with a resolution

better than 5 kyr [from Grobe et al., 1990; Labeyrie and Duplessy, 1985;

Charles and Fairbanks, 1990; Duplessy et al., 1988]. Core RC 11-120 is added for reference [Martinson et al., 1987]. The main frontal zones are reported from Gordon et al. [1982].

phases are associated with large decreases in salinity. Climatic

changes in this area of the Southern Ocean are therefore marked with north-south shifts of the zone between warm, saline,

surface tropical waters and cold, low salinity, polar waters.

They are probably

associated

with similar

shifts

in the westerly

wind belt.

This evolution points to a major role of the low-latitude surface water temperatures and atmospheric circulation as an interhemispheric link during glacial-interglacial transitions.

2. The Isotopic Records of Planktonic

and Benthic Foraminifera

The foraminiferal

180/160

and 13C/12C

isotopic ratios

(expressed

as 8•80 and

813C

versus

Pee Dee Belemnite

(PDB))

have been measured on an automated carbonate preparation line coupled to a Finnigan MAT 251 mass spectrometer. The mean external reproducibility of powdered carbonate standards is

_+0.05%0

for oxygen and carbon.

Data are reported

using as

reference NBS19 8180 of-2.20%0 and 8•3C of +1.95%c versus

PDB [Hut, 1987; Coplen, 1988]. Prior to isotopic analysis, hand-picked foraminifera are cleaned in pure methanol and

heated under vacuum at 400øC.

Both planktonic foraminifera species Globigerina bulloides and Neogloboquadrina pachyderma s. were analysed from the

same size fraction (200-250 pm). Measured •80 values are similar for both species in sections where they are both abun- dant (mean difference between G. bulloides and N. pachyderma

s. is A8•80

= 0.11_+

0.26 %0 for these

periods).

This is not true

during major climatic transitions (Table 1). Bard et al. [1987]

have shown that benthic bioturbation redistributes a fraction of

the shells from depths of high relative abundance to levels above and below. A fraction of the abundant N. pachyderma s.

shells from glacial levels are therefore transported (with their

"heavy"

8•80) to the adjacent

interglacial

levels, where

this

species is less abundant. Conversely, some of the G. bulloides

shells

with their "light" 8180

values,

abundant

in interglacial

levels, are moved to the adjacent glacial levels. This results in

shifting the transition between "warm" and "cold" 8•80 values toward the interglacial for N. pachyderma s. and toward glacial for G. bulloides. Since the present study concentrates on the

timing of climatic changes, this effect must be taken into

account.

Therefore

the mean

planktonic

8•80 signal is built

from both G. bulloides

and N. pachyderma

s. 8•80 data

sets

after

rejection of the values corresponding to a low relative abundance (less than 10%) for the corresponding foraminiferal

species.

Table 1. Isotopic

Data Versus

Depth,

Corrected

for Specific

Fractionation,

for the Top 108 cm

Depth, G. bull. N. pach. C. wull. E. exi. M. barl benth. C. wull. M. barl E. exi. benth.

cm 8•80, %0 8180, %0 8180, %0 8180, %0 8180, %0 8180, %0 813C, %0 813C, e/co 813C, %0 •513C, C7cc 0 1.84' 3.48 3.45 3.31 3.41 0.17 2 1.81 ... 4 1.78 ... 6 2.05 ... 8 1.81 ... 10 - 1.7' 3.37 - - 3.37 0.05 12 1.69 ... 14 1.91 ... 16 1.9 ... 0.28 0.07 0.17 0.05 _ _ _

(4)

LABEYRIE ET AL.: SOUTHERN OCEAN PALEOHYDROGRAPHY 59

Table 1. (continued)

Del•th, G. bull. C. wull. E. exi. M. barl benth. C. wull. M. barl E. exi. benth.

cm /5180, %c •j180, %t• •j180, %t• •j180, %t• •j180, %t• •j180, %c •j13C, %c •j13C, %c •jlaC, %c •jl•C, %c , , , 18 1.96 ... 20 - 1.67' 3.49 3.44 - 3.47 -0.02 - 0.08 0.03 22 1.67 ... 24 1.77 ... 26 1.81 ... 28 1.58 ... 30 - 1.72' 3.49 - - 3.49 0.12 - - 0.12 32 1.59 ... 40 - 1.81 3.93 3.99 3.93 3.95 -0.18 0.44 -0.22 0.01 42 1.54 ... 44 1.98 ... 46 2.08 ... 48 2.35 ... 50 - 2.54 4.12 - 4.43 4.27 -0.02 -0.21 - -0.12 52 2.77 ... 54 2.43 ... 56 2.51 ... 58 2.44 ... 60 - 2.72 4.75 - 4.86 4.81 -0.35 -0.51 - -0.43 62 2.62 ... 64 2.51 ... 66 2.86 ... 68 2.39* ... 70 - 3.65 - - 5.08 5.08 - -0.78 - -0.78 72 2.76* ... 74 2.54* ... 76 2.85* ... 78 2.31' ... 80 - 3.94 - - 5.03 5.03 - 1.04 - 1.29 - - 1.17 82 2.8* ... 84 2.86* ... 86 2.52* ... 88 2.99* ... 90 - 3.7 - - 5.08 5.08 -1.08 -1.12 - -1.1 98 3.37* ... 100 - 4.06 - - 4.86 4.86 -1.22 - -1.22 102 3.06* ... 104 2.53* ... 106 3.26* ... 108 3.41' ...

The

'•"'80

of G. bulloides,

N. pachyderma

s.,

C. wuellersto•fi

(+0.64%,),

E. exigua

(+0.22%c),

M.

barleeanum (+0.4%,), and mean benthic values are given, as well as the bl3C of C. wuellersto•fi, M. barleeanum (+0.83%c), E. exigua (+0.6%,), and mean benthic values.

* Values that are not considered in figures and calculations, because of low foraminiferal abundance.

The benthic

b180

record

(Figure

2) is based

on analysis

of the

individual species Cibicidoides wuellerstorfi, Melonis barlee-

anum, and Epistominella exigua; 64 levels were analysed for more than one benthic species. Results are reported with the

admitted corrections for specific fractionation: +0.64%0 for C. wuellerstorfi and +0.40%0 for M. barleeanum [Duplessy et al., 1984]. A correction of +0.22%0 is derived for E. exigua from

paired measurements of E. exigua and C. wuellerstorfi (mean

difference for the 20 levels sampled during stable paleoclimatic conditions: +0.42 +0.12%o). Grobe et al. [1990] propose that this species grows its shell near isotopic equilibrium but do not

give specific arguments.

C. wuellerstorfi are mostly absent during glacials and M.

barleeanum during interglacials. E. exigua was sufficiently

abundant to be sampled in the interval between 710 and 970 cm (end of isotopic stage 5.5 to 4). As for planktonic foraminifera,

we did not take keep measurements corresponding to a

minimum in relative abundance of a given species.

The mean

benthic

foraminifera

•13C

record

(Figure

2) is built

following the same rules, with corrections derived from paired measurements (0%0 for C. wuellerstorfi, +0.60%0 for E. exigua, and +0.83%0 for M. barleeanum).

The final foraminiferal /5•80 and /5•3C data set is available from L. Labeyrie, and will be transferred to the Boulder World Data Center-A (WDC-A) after publication. A partial example (upper 108 cm) is given in Table 1.

3. Micropaleontology

3.1. Diatoms

Diatom species distribution has been studied at 5- to 10-cm intervals. Changes in sea surface temperature were reconstructed using the diatom transfer function T.F.166/34/4 [Pichon et al.,

1987, 1992]. Its construction follows Imbrie and Kipp [1971], except that the relative amounts of diatoms of each of the 34

(5)

60 LABEYRIE ET AL.: SOUTHERN OCEAN PALEOHYDROGRAPHY 10 .• 8

E 4

E 0 400 800 1200 1600

depth in core MD 88-770

( cm )

Figure 2. Variations with depth in the core of (a) planktonic foraminiferal •5•SO; (b) benthic foraminiferal

•5•80;

(c) summer

sea surface

temperature

(SSST)

calculated

using

the diatom

transfer

function'

(d) SSST

estimates using the foraminiferal transfer function. Isotopic stages are defined following Imbrie et al. [1984].

species considered are scaled to the mean relative abundance of

that species in the reference database. The scaling is done using

four classes of abundance (absent, less than 2%, between 2% and half of maximum abundance for that species, more than half of the maximum abundance). This method increases the sensitivity of the transfer function to changes in low abundance

species [Pichon et al., 1992] which have well defined tempe-

rature tolerance and are thus important for the transfer function.

Calibration is done on a set of 124 Holocene and modern core

tops. Forty-two samples with assemblages altered by stepwise

opal dissolution of five Holocene core tops were added to the

natural flora data to take into account the effect of dissolution.

The transfer function predicts correct summer surface water temperature in the range -1 to +12øC with a mean standard

deviation (1 c•)of +I.IøC, when tested for modern core tops [Pichon et al., 1987, 1992].

The first three factors of the factor analysis (varimax matrix), which represent 34%, 30.2% and 12.6 %, respectively, of the total variance, correspond to three typical biogeographic associations. They group the open ocean species (Thalassiothrix sp., Nitzschia kerguelensis, and the silicofla- gellate Distephanus speculum), the ice-related species (Thalas- siosira antarctica, Actinocyclus actinochilus, Nitzschia sublineata, Nitzschia cylindrus and Eucampia antarctica) and the

Subantarctic species (Thalassionema nitzschioides, Roperia tesselata, Hemidiscus cuneiformis, and the silicoflagellate Dictyocha sp.), respectively. The fourth factor is directly

(6)

core MD 88-770, the ice-related factor represents only a small part of the variance, even during cold periods. This suggests that the core was always between the subtropical and polar fronts. However, summer sea surface temperature estimates during glacials, around +2øC, do indicate that the polar front was not far south during these periods.

3.2. Planktonic Foraminifera

The foraminiferal species distribution has been established at 5- to 10-cm intervals along the core and at higher resolution for the upper 4 m (2-cm intervals). Construction of the transfer function follows also Imbrie and Kipp [1971]. It is built on the modern reference database of Prell [1985], with an extended core top reference at high southern latitudes [Labracherie et al., 1989]. SSST temperatures are interpolated from Levitus's Atlas mean February to April 0- to 10-m water temperatures [Levitus, 1982]. The foraminiferal transfer function, when tested on individual core tops against the modern reference database, gives SSST estimates between 0.7 and 23øC with a c5 = 1. IøC. However, the precision is much poorer in the SSST range -1 to +3øC than above, as the foraminiferal population is then essen- tially monospecific N. pachyderma s. Details of the micropa- laeontological study will be discussed in a paper in preparation by M. Labracherie et al. (manuscript in preparation, 1995).

The foraminiferal population varies between two characteristic assemblages in the core: the first, which corres- ponds to the warm periods, is typical of subpolar waters (with G. bulloides and G. infiata); the second, during the cold periods, is composed of the quasi-monospecific N. pachyderma s. population (which corresponds to a maximum SSST value of +3øC). Such an evolution is in agreement with the results of the diatom study. Results of both transfer functions (Figure 2) will be available from M. Labracherie and deposited at the Boulder World Data Center-A after publication. A partial example of the data is reported as Table 2.

3.3. Discussion of the Micropalaeontological

Results

Both transfer functions give similar signals. The only signi- ficant differences are for the low- and high-temperature extrema.

Let us consider first the interglacial periods. During the Holocene (the top 100 cm)and the warm interstadials of isotopic stages 5.1 and 5.3 (about 800 and 900 cm respectively), the diatom transfer function gives SSST (Figure 2 c) about 2øC higher than the foraminiferal SSST (Figure 2d), i.e., +9-10øC versus +7-8øC. The modern SSST given by Levitus [1982] at the core location is +9øC. The foraminiferal estimate appears therefore systematically too cold by about IøC and the diatom estimate too warm by that amount for the core top. At the opposite, during the peak interglacial of stage 5.5 (about 1000 cm), SSST are higher for foraminifera (10.5øC versus 8.5øC for diatoms). The peak warmth, at 1004 cm (+13øC, foraminiferal SSST), has not been sampled for diatoms but would fall out of the calibration range of the diatom transfer function. Core RC 11-120, located 2030 ' further north, in sub- tropical waters (Figure 1), yields SSST estimates of +11.2øC for Holocene and +13.5øC for the 5.5 interglacial (Spectral Mapping (SPECMAP) file, WDC-A database). These values are compatible with both sets of values for core MD 88-770, taking into account the latitudinal difference.

We do not have ready explanations for the observed dif-

ferences between diatom and foraminiferal transfer functions.

They appear systematic within each climatic period. They may derive from changes in the upper water structure and seasonality of the productivity or from a contribution of shells transported by the circumpolar currents from more southerly or northerly environments. In the latter case, the diatom transfer function

would be much more affected than the foraminifera transfer

function, as it is more sensitive to relative changes in low abundance species. During cold periods, diatoms give mean

SSST values about 1 øC colder than foraminifera. The difference

exceeds 2ø-3øC during the maximum cold events of stages 2, 4, and 6. Such a difference may be expected. As opposed to fora- minifera, diatoms are far from their minimum temperature range and will grow even during the colder years. Contrary to forami- nifera, they have a high species diversity, and the diatom

transfer function is well validated to -IøC SSST. In conclusion, records obtained from diatom and foraminiferal transfer

functions are not completely equivalent. The foraminiferal transfer function appears better adapted to record the evolution of the subtropical front in relation with the foraminiferal iso- topic record. It has no meanings for SSST below approximati- vely +3øC. The diatom transfer function presents a specific interest for the low-temperature range (polar waters) and for recording sea ice and related hydrology.

Another feature of the records is the apparent lead of forami- nifera versus diatom, particularly evident in the early part for the last deglaciation. This point will be considered in the

discussion.

4. Data Analysis and Discussion

4.1. Chronostratigraphy of MD 88-770

The different isotopic stages 1 to 7 may be easily recognised in the isotopic records. However, it is evident from Figure 2 that sedimentation rates were significantly higher during

glacial periods. Chronology for the paleoceanographic records

from core MD 88-770 was developed in the following steps.

First, accelerator

mass

spectrometry

(AMS) •4C dating

of N.

pachyderma s. was used to build a preliminary timescale for the

upper 500 cm. Dating followed the standard procedures at Centre des Faibles Radioactivit6s (CFR), Gif sur Yvette, with

sampling limited to peaks in relative abundance of the considered species. This minimises the effect of bioturbation.

Two levels present anomalous young ages (554 and 418 cm depth). We suspect contamination during cleaning. These

levels were excluded from the construction of the timescale.

AMS 14C

ages are reported

(Table 3 and Figure

3) after correc-

tion by 480 years, the adopted mean reservoir age [Bard et al.,

1990] We did not adjust the reservoir corrected ages to get calendar ages, as proposed by Bard et al. [1993]. The present

calibration

of the •4C scale

is far from being accurate

for ages

older that 11 kyr B.P. [Bard et al., 1993].

Second, the benthic foraminiferal •5•80 record of the lower

1170 cm was graphically correlated with the benthic record of

equatorial Pacific core V19-30 [Shackleton and Pisias, 1985],

using the Analyseties software developed at CFR (D. Paillard et al., CFR internal report, 1993). We chose the benthic foramini- feral/5•80 record of core V19-30 as SPECMAP reference for its

close

similarity

even for small

details

with the MD 88-770 •5•80

(7)

62 LABEYRIE ET AL.: SOUTHERN OCEAN PALEOHYDROGRAPHY

Table 2. Micropaleontological Data Versus Depth in the Core for the Top 94 cm and Derived Summer Sea Surface Temperature (SSST) Following Imbrie and Kipp [1971]

Diatoms Foraminifera

Del•th, cm Ol•en Ocean Ice Zone Subantarctic Com. SSST, øC pach. s., % bull., % Com. SSST, øC

0 0.49 0.13 0.76 0.85 9.5 16 40 0.99 7.8 2 ... 16 48 1.00 7.0 4 ... 13 52 0.99 7.8 6 ... 15 50 1.00 7.3 8 ... 12 50 1.00 7.8 10 0.09 0.77 0.83 10.5 17 44 0.99 7.1 12 ... 12 48 1.00 7.9 14 ... 15 45 0.98 7.5 16 ... 16 44 0.99 7.4 ! 8 ... 21 42 0.99 7.1 20 0.44 0.12 0.78 0.85 ! 0.3 19 40 0.98 7.3 22 ... 15 43 0.98 7.6 24 ... 16 46 1.00 7.1 26 ... 17 48 0.99 7.1 28 ... 13 49 0.99 7.5 30 0.53 0.14 0.73 0.86 9.0 9 56 1.00 8.7 32 ... 13 53 0.99 7.8 34 ... 14 50 0.99 7.4 36 ... 18 42 0.99 7.4 38 ... 14 51 0.99 7.3 40 0.54 0.13 0.71 0.89 9.2 16 48 0.99 7.1 42 ... 13 45 0.98 7.5 44 ... 16 45 0.98 7.4 46 .... 19 46 0.99 6.8 48 ... 21 45 0.99 6.9 50 0.49 0.09 0.72 0.82 10.3 14 53 0.99 7.8 52 .... 18 50 0.99 7.0 54 .... 20 55 0.99 7.2 56 .... 20 52 0.97 7.5 58 ... 19 51 0.98 7.2 60 0.51 0.26 0.68 0.79 7.0 21 51 0.98 7.0 62 .... 26 50 0.99 6.6 64 .... 21 53 0.99 6.9 66 .... 21 49 0.97 7.2 68 ... 16 56 0.98 7.7 70 0.66 0.26 0.46 0.74 5.1 22 56 0.98 7.0 72 ... 23 51 0.99 6.7 74 ... 23 40 0.94 7.5 76 ... 40 34 0.97 7.2 78 ... 54 27 1.00 6.4 80 0.72 0.32 0.29 0.72 3.1 33 43 0.99 6.7 82 .... 44 30 0.99 7.2 84 .... 33 40 0.99 6.7 86 .... 40 38 0.99 6.9 88 .... 75 14 1.00 4.4 90 0.72 0.35 0.32 0.76 2.8 30 49 0.98 6.8 92 .... 82 10 1.00 3.9 94 ... 92 4 1.00 3.1

Results

of the factor analysis

of the diatom

population

were derived

following

Pichon et al. [1992]. The relative weight of

the Open Ocean, Ice Zone, and Subantarctic diatom associations, their communality (Com.) and the calculated Summer Sea

surface temperatures (SSSST) are given, as well as the relative abundance of the planktonic foraminifera N. pachvderma s.

(polar factor) (pach.s),

and G. bulloides

(transitional

factor) (bull.), their communality,

and SSST, calculat4d

by the

foraminiferal transfer function.

record (Figure 3). This helped to accurately define the times of rapid change in sedimentation rate. V 19-30 has been used in the "short stack" benthic record of Pisias et al. [1984] and is defined in the temporal SPECMAP framework of Martinson et al. [1987].

Finally, the timescale derived from the two first steps was

further modified using Martinson et al.'s inverse correlation technique (still with the benthic record of V19-30) [Martinson

et al., 1982]. This step improves coherency in the cross corre- lation between the MD 88-770 isotopic record and insolation and puts the corresponding palaeoenvironmental changes within the SPECMAP temporal framework [Imbrie et al., 1992, 1993]. The poor control at both ends of the record introduces a mean shift of +0.7 kyr (+_1 kyr at 1 •)when compared to the graphically derived timescale. The effect is particularly apparent for the upper 600 cm (the last 50 kyr) and the

(8)

Table 3. Results

of the Accelerator

Mass Spectrometry

x4C

dating of N. pachyderma s.

Depth, cm Age, ka Corrected Ages Error at 1 Sigma

20 6.18 5.70 0.10 46 11.48 11.00 0.12 58 12.63 12.15 68 12.01 11.53 0.15 74 13.98 13.50 0.15 114 17.26 16.78 0.22 228 22.09 21.61 0.30 256 20.88 20.40 0.26 278 21.77 21.29 0.29 304 25.64 25.16 0.39 342 25.49 25.01 0.45 360 27.86 27.38 0.51 384 30.98 30.50 0.67 436 33.27 32.79 0.82 472 37.70 37.22 1.50 524 45.10 44.62 3.30 418' 27.31 26.83 0.51 554* 37.20 36.72 1.30 * values rejected

transition from isotopic stages 6 to 5.5, where ages increase by 1.5 to 2 kyr. This brings the calculated ages for the 11-40 ka period close to what would have been estimated using the correction for calendar ages [Bard et al., 1993]. The global

correlation coefficient of the dated benthic records is 0.96

between MD 88-770 and V19-30 (Figure 3) and 0.91 between MD 88-770 and the SPECMAP short stack, respectively. Final depth/age results for the upper 416 cm are reported in Table 4 for example. This data will also be available from the Boulder

World Data Center-A.

4.2. Evolution Over the Last 230 kyr

The different records are reported versus age in the core in Figure 4. As discussed previously, the SSST estimated by the foraminiferal transfer function is considered to better represent the overall variability of surface waters than the SST estimated by the diatom transfer function, although it does not record temperature excursions below 3øC (Figure 4a). SSST presents a rapid variability for both glacials and interglacials approxima-

tively in phase

with the planktonic

foraminiferal

8•80 record

(Figure 4b). The latter signal records the effects of both the

changes

in SSST

and ocean

water

8x80. The largest difference

between

SSST and planktonic

8i80 signals

occurs

during

isoto-

pic stage

5.5, where

the second

light 8•80 peak

is not paralleled

by a SSST increase.

The benthic foraminiferal

8•80 signal

(Figure 4c) records the continental ice sheets waxing and waning but also the effect of deep water temperature changes [Shackleton, 1967; Labeyrie et al., 1987]. A lead of SSST and

planktonic

foraminiferal

8•80 over benthic

8•80 appears

during

each cooling phase. The benthic foraminiferal

8•3C record

(Figure 4d), although rather noisy, shows strong analogies with the benthic 8•O record. As for other Southern Ocean cores [Duplessy et al., 1988; Curry et al., 1988; Charles and

Fairbanks,

1992], exceptionally

light 8•3C values

are observed

during glacial periods, compared to records from the other oceanic basins. The interpretation of that signal is still a

matter of debate, but there are strong indications that, as in the

modem

ocean, light 8•3C values

record

accumulation

of •3C

poor dissolved CO: (produced by oxidation of settling organic matter) in the deep southern waters following the isolation of this water mass during glacial periods [Michel et al., 1995, and references therein]. Whatever the precise origin of that signal, we may safely admit that a return, even partial, to the modern thermohaline circulation, with ventilation of the deep Southern Ocean by NADW and/or Antarctic Bottom Water (AABW), will shift the benthic foraminiferal •t3C to values closer to those of the deep Pacific and North Atlantic records [Charles and Fairbanks, 1992]. In that respect, information carried by the benthic foraminiferal 15•80 and 15•3C is different: 15x80 tracks the

northern ice sheets and their influence on NADW; 15•C tracks

the isolation of Southern Ocean deep waters. 4.3. Study in the Spectral Domain

Spectral analysis of the climatic response to the insolation

forcing may help to clarify the relationships between Southern Ocean climatic records of core MD 88-770 and global climate. Imbrie et al. [1992, 1993] have shown that the periodic compo- nent of paleoclimatic proxies variability may be considered as a sum of three independent elements, the response to preces-

sion (=1/21 kyr-•), obliquity

(=1/43 kyr'X),

and continental

ice

volume

(which

contributes

largely

to the 1/100 kyr -• frequency

band). This third parameter is probably itself a nonlinear

response to precession, obliquity, eccentricity, and their

harmonics. However, seen from the Southern Ocean, we may

consider the continental northern ice sheets as an external

forcing

acting

primarily

in the 1/100 kyr -• band.

Paleoclimatic

response to insolation appears linear in the precession and obliquity bands [Imbrie et al., 1992, 1993, and references

therein], but both bands carry different information. Mcintyre et al. [1989], among others, have shown that low-latitude paleoclimatic responses (sea surface temperature, equatorial upwelling) have a major part of their variability in the preces- sion band. At the opposite, proxies which record processes linked with high-latitude climatology (benthic foraminifera

15•80

and 15•C, Cd/Ca, lysocline depth) have most of their

variance

in the 1/100- and 1/41- kyr '• bands [Imbrie et al.,

1992, 1993, and references therein].

Each of the climatic bands has its own properties when considering interhemispheric relationship:

The

first

is the 1/100-

kyr

-1 "ice

forcing"

band.

Most

of the

continental glacial-interglacial ice volume changes occur in the northern hemisphere [CLIMAP, 1981]. Since purely low- latitude paleoclimatic proxies do not exhibit strong variability

in the 1/100 kyr

'1 band

[Mch•tyre

et al., 1989], ice sheet

forcing on southern latitudes acts probably through a modula-

tion of the deep water dynamic and chemistry and its resulting effect on heat transfer and atmospheric pCO 2.

The second is the 1/43 kyr -• "obliquity" band. This

parameter modulates the distribution of solar energy between low and high latitudes synchronously over both hemispheres.

The third is the 1/21 kyr 'x "precession"

band. Insolation is

modulated by the relative positions of perihelion and summer solstice. It is in opposition of phase for both hemispheres. This is the major band where interhemispheric transfer processes through atmosphere and surface oceans should play a significant role.

(9)

64 LABEYRIE ET AL.: SOUTHERN OCEAN PALEOHYDROGRAPHY 1800 400 ooo 600 200 MD 88-770 I I I _t _t _I _I _I _I _I _I _I _I _I _I _I _I _I _I _i

0

50

100

150

200

250 age in the SPECMAP

scale ( ka )

i i i i i

I-I AMS

14 C dates

(planktonic

foraminifera),

m rejected

0 SPECMAP

tie-points

/ final age model

Figure 3. Construction

of the timescale:

(a)Age-depth

relationship,

with accelerator

mass

spectrometry

(AMS) 14C

dates,

tie points derived

from the graphical

correlation

of MD 88-770 and V 19-30 benthic

foraminiferal

6180

records,

and

the final inverse

correlation

using

Martinson

et al. [1982] method;

(b) The

(10)

Table 4. Depth versus Age Scale fbr the upper 416 cm of Core MD 88-770

Depth, cm ka Depth, cm ka Depth, cm ka Depth, cm ka

0 4.59 104 17.23 208 21.62 312 26.46 2 4.87 106 17.39 210 21.69 314 26.58 4 5.16 108 17.55 212 21.77 316 26.70 6 5.44 110 17.70 214 21.84 318 26.83 8 5.72 112 17.86 216 21.92 320 26.95 10 6.00 114 18.02 218 21.99 322 27.08

12

6.28

116

18.10

220

22.07

324

27.20

14 6.57 118 18.17 222 22.15 326 27.32 16 6.85 120 18.25 224 22.22 328 27.45 18 7.13 122 18.33 226 22.29 330 27.57 20 7.41 124 18.41 228 22.38 332 27.69 334 27.81 22 7.71 126 18.49 230 22.46 336 27.93 24 8.02 128 18.56 232 22.55 338 28.06 26 8.32 130 18.64 234 22.63 340 28.18 28 8.62 132 18.72 236 22.72 342 28.31 30 8.92 134 18.80 238 22.81 344 28.44 32 9.22 136 18.88 240 22.90 346 28.58 34 9.52 138 18.95 242 22.99 348 28.73 36 9.82 140 19.03 244 23.08 350 28.87 38 10.12 142 19.11 246 23.17 352 29.01 40 10.42 144 19.18 248 23.25 354 29.15 42 10.72 146 19.26 250 23.34 356 29.29 44 11.02 148 19.33 252 23.43 358 29.44 46 11.31 150 19.41 254 23.52 360 29.58 48 11.61 152 19.49 256 23.61 362 29.73 50 11.91 154 19.56 258 23.69 364 29.87 52 12.20 156 19.64 260 23.78 366 30.01 54 12.50 158 19.72 262 23.87 368 30.15 56 12.79 160 19.79 264 23.96 370 30.29 58 i3.08 162 19.87 266 24.05 372 30.43 60 13.33 164 19.95 268 24.13 374 30.57

62

13.58

166

20.02

270

24.22

376

30.72

64

13.83

168

20.10

272

24.31

378

30.86

66

14.07

170

20.18

274

24.40

380

31.00

68

14.32

172

20.26

276

24.49

382

31.14

70

14.48

174

20.33

278

24.59

384

31.28

72

14.64

176

20.41

280

24.68

386

31.45

74

14.81

178

20.48

282

24.79

388

31.62

76

14.97

180

20.56

284

24.90

390

31.79

78

15.13

182

20.64

286

25.01

392

31.96

80 15.30 184 20.71 288 25.11 394 32.13 82 15.46 186 20.79 290 25.22 396 32.30 84 15.62 188 20.86 292 25.33 398 32.47 86 15.78 190 20.94 294 25.43 400 32.64 88 15.94 192 21.02 296 25.54 402 32.81 90 16.11 194 21.09 298 25.64 404 32.98 92 16.27 196 21.16 300 25.75 406 33.14 94 16.43 198 21.24 302 25.86 408 33.31 96 16.59 200 21.31 304 25.97 410 33.48 98 16.75 202 21.39 306 26.09 412 33.65 100 16.91 204 21.47 308 26.22 414 33.82 102 17.07 206 21.54 310 26.34 416 33.99

In order to perform spectral analysis, foraminiferal isotopic and SST temporal records (Figure 4) were interpolated at a 0.5-kyr interval between 4.5 and 235.5 ka. Spectral analysis was done by the Blackman-Tukey method [Jenkins and Watts, 1968] using the software Analyseries. The Blakman-Tukey spectral analysis algorithm is derived from the one used in Brown University within the SPECMAP program [Imbrie, 1985; Imbrie et al., 1989]. The analysis is performed with a

Barierr-type

window,

(220 lags and a 0.024 kyr -I bandwidth).

For easier intercomparison of the different parameters, variance powers are normalised by the integrated variance over the

studied

frequency

range (0 to 0.1 kyr4). They are therefore

presented as relative variance powers. The five studied. parameters, i.e., SSST (diatom and foraminifera), foraminiferal

•5•80

(benthic

and planktonic),

and •5t3C

(benthic) present

the

same general spectral responses (Figure 5), with variance

maxima

around

1/106 kyr '•, 1/43 kyr '•, 1/23 kyr

'•, and 1/19

kyr -•. Additional

high-frequency

harmonic

responses

around

1/13.5

and

1/10 kyr-1, which

represent

less

than

2% of the

integrated variance, are not reported. In the following

discussion,

considerations

about the 1/100-kyr

-• band should

be considered as semiquantitative. The 240-kyr record is too short for results to be fully reliable in that frequency band.

(11)

66 LABEYRIE ET AL.: SOLr•ERN OCEAN PALEOHYDROGRAPHY

isotopic stages

o

0 50 1 O0 1 50 200 250

SPECIVlAP

time scale ( ka )

Figure 4. Baleoproxies

from

core

MD 88-770

versus

a•e (ka)' (a) summer

sea

surface

temperature

(SSST)

using

the foraminiferal

transfer

function;

(b) planktonic

foraminiferal

15•80;

(c) benthic

foraminiferal

15•SO;

(d)

benthic foraminiferal 15•'•C.

Following lmbrie et al. [1992], cross correlations (Blackman-Tukey) are calculated between the different Southern

Ocean

climatic

proxies,

the benthic

15•80,

and the northern

hemisphere insolation (June 15 at 65øN)records for the three major bands of the climatic response at 1/106, 1/43, and 1/23

kyr

-• (Figure

5 and

Table

5). Covariances

with the benthic

15180

record are highly significant (0.75<Coh<0.99) for the three bands and all the proxies (80% confidence interval for arctanh (Coh)=+0.52 [Jenkins and Watts, 1968]). Covariance between

the different

Southern

Ocean

proxies

is not significant

for the

1/106

kyr

'• band

(0.23<Coh<0.31)

and is significant,

but with

a lower coherence than between the benthic 15•SO record and

(12)

LABEYRIE ET AL.: SOUTHERN OCEAN PALEOHYDROGRAPHY 67 1.2 0.8 0.4 0.4 Benthic •18 0 1- Foraminiferal T.F. 2- Diatoms T.F.

3-

Planktc•pic

õ18

0

4 Ben•hic 6?80 5 Benthic 6 • 3C Insolation

1-

Foraminiferal

_I

T.F.

0- 5- _ ' 5 ,

2-

D•atoms

T.F.

_ 4 Benthic 8180 -5 Benthic 6 3 3C 0 0.01 0.02 0.03 0.04 O.OS 0.06 0 0.01 0.02 0 03 0 04 0 05 0 06 frequency ( 1/kyr) frequency ( 1/kyr)

Figure 5. Results of the spectral analysis (Blackman-Tuckey) and cross correlation with the (left) benthie

6•80

record

and

(right)

insolation

for (t¾om

top

to bottom)

foraminiferal

SSST,

diatom

SSST,

planktonic

6•O,

benthie

•80, and

benthie

•3C. Left scales

show

distribution

of the relative

variance

power.

Right

scales

show

phases (in radians) and coherency. F.T. is transfer function.

(13)

68 LABEYRIE ET AL.' SOUTHERN OCEAN PALEOHYDROGRAPHY i i i • • i i i i i i i i i i i i i i i i i i i i i i 0 0 0 0

• .•

o •

u

u .-

(14)

LABEYRIE ET AL.: SOU22-IERN OCEAN PALEOHYDROGRAPHY 69

insolation,

in the 1/43-kyr

-• (0.56<Coh<0.79)

and 1/21-kyr

-•

(0.71<Coh<0.88) bands. This is in agreement with the results of Imbrie et al. [1992, 1993]. The large coherent response of

the various

parameters

with ice volume

at 1/106 kyr

-1 is directly

linked to the global influence of the large northern ice sheets, which have paced the climatic records at that frequency over

most of the last 900 kyr. The response of the Southern Ocean

proxies

is also larger

in the 1/43-kyr

-• that in the 1/21-kyr

'•

bands. This is related to a recent study by Waelbroeck et al.

[1995], in which the Vostok deuterium fid record (a proxy of

the air temperature over Antarctica) is extended to 250 ka and

analyseo in the SPECMAP temporal framework, by direct

tuning

with insolation

in the 1/43- and 1/20-kyr

-• bands.

Their

study show that the Vostok fid record has a larger fraction of

the variability

in the 1/43-kyr

-• band

than in the 1/100- and the

1/20-kyr

-t bands,

when compared

to the marine

series. This

points to a direct effect of high-latitude southern insolation on the local changes in air and sea surface temperature.

When

we consider

the leads

and

lags

between

potentiai

forcings and responses, the first important result is the systematic lead (1 to 3 kyr) of the planktonic versus the

benthic

foraminiferal

fi•80 records

in the three

frequency

bands.

Thus the choice of referring chronostratigraphy to benthic or planktonic foraminiferal records (as in SPECMAP files) has significant implications in the Southern Ocean. SSST leads the

benthic

foraminiferal

fi180

record

by about

3 kyr for the diatom

transfer

function

and 3 kyr (at 1/43 kyr -•) to 7 kyr (at 1/21

kyr -•) for the foraminiferal

transfer

function.

This SSST

lead is

reduced to about 1-2 kyr when compared to the MD 88-770

planktonic

fi•80 record,

in agreement

with previously

published

results for RC11-120 and other Southern Ocean cores [Imbrie et al., 1992; 1993; Howard and Prell, 1992]. This result is not

surprising

since the planktonic

foraminiferal

fi•80 signal is

driven partly by SSST (the leading signal) and partly by global

seawater

fi•80 (the lagging

component),

recorded

by the benthic

foraminiferal

fi•80.

Our results

imply that northern

hemisphere

ice sheets may participate but are not at the origin of the

changes in Southern Ocean SSST in the obliquity and

precession bands.

SSST (foraminiferal transfer function) lags maximum obliquity by 3.5 kyr (31ø). This points to climatic processes

with a time constant of several kiloyears, as the evolution of the peripheric extent of ice sheets or deep water formation and

transport [Waelbroeck et al., 1995]. In the precession band, the

foraminiferal SSST record lags the June 15 summer insolation

by 2 kyr (17ø). This corresponds to a change in Southern Ocean summer surface temperature in phase with July insolation

(northern summer).

The diatom transfer function has a significantly different

response, considering the + 1- to 2-kyr phase uncertainty calculated by the SPECMAP method (Table 5): it is approxima- tively in phase with insolation in the obliquity band but lags

insolation by 6.5 kyr in the precession band. This is to be related with the better sensitivity of the diatom transfer function to the cold polar water and sea ice variabilities. In that respect, it is acceptable that high-latitude southern surface waters are very sensitive to the direct influence of the obliquity

modulation of seasonality. The larger lag in the precession

band points also to a higher sensitivity of the diatom distribution to the local climatic influence of high-latitude

processes with large time constant as the Antarctic ice sheet peripheral extent.

This still does not explain why the foraminiferal SSST signal, although less sensitive to the high-latitude processes,

presents a larger lag in the obliquity band than the diatom SSST

signal. If the results are taken at face value, they indicate that at least two processes influence the Southern Ocean climatology in the obliquity band: (1) locally, at high latitude, the sea ice coverage and SSST which react with no lags to changes in insolation; (2) at lower latitudes, the ice sheet peripheral extent and related atmospheric and deep water forming processes. In

reverse, in the precessioh band, the low-latitude insolation forcing would be the predominant factor for the foraminiferal SSST signal.

The phasing of the benthic foraminiferal

•5t3C

response

helps to define precisely the role of deep waters. The fi•3C leads

the benthic

fi•80 signal

by 1 to 2.5 kyr in both orbital bands

but lags SSST by 2 to 5 kyr. Our results confirm earlier work by ?isias and Shackleton [1984] in the deep equatorial Pacific and by Charles and Fairbanks [ 1992] on the last deglaciation in the deep Southern Ocean: dissolved CO2 and its transfer to the atmosphere and continental biosphere occured early in the chain of events which drove to interglacials. But we demons-

trated that surface water temperature changes occured before. The key to the early evolution of the Southern Ocean surface

waters is therefore not to be found in deep water and its carbon

cycle. The rather

large

phase

of the benthic

15t3C

response

to

insolation,

about

5 kyr for the 1/43-kyr

-t and 1/21-kyr

'• bands,

is compatible with our interpretation of the diatom and

foraminiferal SST response: it is similar to the "slow" response mode of the foraminifera in the obliquity band and of the

diatoms in the precession band which we have attributed to the peripheral evolution of the Antarctic ice sheet and its influence on atmospheric circulation and deep water formation.

4.4. The Salinity Hypothesis

In the northern hemisphere, one of the main characteristics of glacial periods is the formation of a frontal zone across the North Atlantic, with warm and saline waters to the south and cold and less saline polar waters to the north [CLIMAP, 1981, 1984; Duplessy et al., 1991, 1992b]. This front is associated with a large decrease of the NADW conveyor belt [Broecker and Denton, 1989]. Both effects efficiently lower the transfer of

salt from the high-salinity intertropical Atlantic to the deep

waters. Labeyrie et al. [1992] have shown that due to the relative salinity increase in warm water and salinity decrease in high-latitude cold surface and deep waters, the density gradients between saline subtropical waters and deep waters were probably much smaller during the last glacial. Therefore zones

of open water convection (modal waters of McCartney [1977])

may have been more generalised near subtropical fronts, changing patterns of heat transfer to the atmosphere and atmospheric circulation.

The circumpolar frontal zone which limits the Southern Ocean to its north is wind driven and strongly linked to zonal sea surface temperature (SST) distribution. We may therefore

consider that similar mechanisms have been operating in the northern and southern hemispheres, with a specific role for the